Bottom line up front: The AI is just not ready for prime time in long games.
I have read the dedicated threads discussing issues with the AI, and read Frogboy’s and others’ defense of the current AI. After playing deeper into an insane map game, I finally understand the criticisms.
Note: While I am certainly no stranger to GC 2 and 3, I am not a hardcore player who can quote XML. The results below were very surprising to me.
Played as Macian (my own custom race, Terran tech tree).
Galaxy Type: Insane, Spiral
Opponents: 26 AI plus all minors
No mods were used.
As of Turn 276
Here is the current galaxy map (which I can see thanks to a single huge hull loaded with nothing but sensors—which I also don’t think should be possible). I tried to outline my territory to make it easier to see what I control.
BUT, the scores tell the real story.
I was really enjoying the game for about 200 turns. I wasn’t the most powerful race and it was a struggle to keep expanding while holding off threats from other races. Now, at turn 276, I’m bored. I know I’m going to win. The rest of the game (barring a mega event that I’m not willing to wait for) is just mopping up. This is not fun.
I can’t lose. Literally. Look at the stats for just one of my medium ships compared to medium ships from three other powers. This is representative of ALL of the AI ships currently in the game. (Oh, I have huge hulls, too. Not one AI even has large hulls yet.)
Notice that not only has the AI not kept pace technologically, but that the ships are not even designed to counter me—the largest threat in the game. I have no beam weapons on any of my ships. As a strategy in this game, I traded for beam weapon tech and then gave that tech to every AI in the game in exchange for other tech or money. The result seems to be that the AI considers beam weapons (which the pirates also use) the primary threat. I used exclusively missile weapons in this game.
So why are my ships (and colonies) so kickass? This one stat tells it all: Research.
So what is going on here? For one, I have research worlds that take maximum advantage of adjacency bonuses. Two, I colonized more worlds than any other player which allows me to have more research-specialized worlds. Three, I trade for tech mercilessly, taking special advantage of the minor races. That allows me to concentrate on research tech and raw production while getting other tech I need through diplomacy.
Lastly, I survey anomalies like crazy. This last one is surprisingly important. With the new costs of “breakthrough” techs such as large hulls, it can take 70+ turns to research. However, with 6-8 survey ships constantly surveying anomalies, I am bound to get +25% research or “Unexpected Genius” every few turns. I got large hulls in 6 turns by surveying anomalies. I got huge hulls in 2 turns. No AI has large hulls. The AI is obviously not capable of following this “short cut” strategy. I also completed the research, production, governing, terraforming, missile, and defense branches with this same strategy.
The Strategic Game
So what would a human player do? Take a look again at the map and my score. Ask the question: If all of the other major players in the game were human players, what would they do? If it were me, I would build a coalition of other players to counter the growing threat. I would coordinate research (you research large hulls and I’ll research more advanced weapons and then we’ll trade). I would work to isolate the major threat diplomatically (in my current game I am financing my heavy spending on research through diplomacy trades). I would seek to create alliances and mutual defense treaties and maybe engage in a coordinated military strike or few until the threatening player had been put in their place. The AI seems incapable of doing any of this, of acting in a coordinated manner for their mutual benefit.
Frogboy and the AI team, I understand (somewhat) the difficulty of the task, having played 4X games since the original Civilization. I appreciate a lot of the behavior I’ve seen in the tactical AI. However, unless the AI can learn to play the strategic game against a human, this just isn’t fun mid- to late game.
- The AI needs to challenge the human player in the colony rush. At turn 276, I’m still sending out colony ships to take worlds that the AIs haven’t claimed within their own borders (extreme southern part of the map).
- The AI needs to learn to work collectively with other AIs against any power (human or AI) that is becoming so powerful as to threaten the collective whole. One other thing I noticed is that no other AIs were gobbling up smaller, weaker AIs to create larger empires that could challenge mine.
- The AI needs to understand how to use the same tricks the human player can use, or the human player shouldn’t be allowed to do it. Adjacency bonus is an obvious game mechanic for this one. So is the anomaly exploit. (I would remove the unexpected genius bonus and lower the +% research as a quick fix.)
- Finally, at least some AIs need to be as ruthless as a human can be (or as loyal). This means breaking non-aggression pacts and alliances, setting other powers to war against or embargo an ally or non-aggression-pact power, etc. Backstabbing is a well-worn human trait. Some AIs should learn it. (Others should be as loyal as the U.K. to the U.S.)
I hope this helps. I REALLY loved GC2 and I want to love GC3 even more. There are many good things about this game. However, for now, I'm going to play the campaign as a change of pace and design some more ships and hope the AI gets a little smarter.